I am approaching to a new project, and have one scene very noisy, I splitted tracks, placed correctly roomtone, smoothed the transition,etc etc, ( applyed many john purcell advises) now i need to reduce the overall noise from the whole scene, how to do that? routing the tracks on an aux track with denoiser inserted to it? or via audiosuite and working on region by region, what is the correct workflow?


Are you mixing the project as well? If not, you should consult the mixer first before you denoise anything, as that is traditionally the mixer's responsibility, and chances are they have their own specific ways they go about it. The lines are getting blurry, though.

My personal workflow is as follows:

1) Do all dialogue editing without any denoising.

2) Duplicate all dialogue tracks (or, alternatively, just duplicate the playlists). I can't emphasize this enough; you never want to do any denoising on your original dialogue tracks. There's nothing worse than listening to your stuff on the dub stage and realizing that you overdid it without any way to go back to the original.

3) Go region by region and apply denoising via AudioSuite (I use iZotope RX). If you set your AS settings correctly ("create individual files" and "clip by clip") Pro Tools will keep all of your fades. Chances are you will have to massage them a bit after denoising, so this step is very important. Also make sure you give yourself enough handles.

Sometimes, if there is just a bit of noise and it's very consistent across setups, you can get away with routing everything to a bus and then doing gentle denoising there, for example with Cedar. However, this only works if you're also the mixer on the project.

One more word of advice: it is very easy to go overboard with denoising, especially at the beginning. Listen very closely, preferably with good headphones, for any artifacts or coloration in the voice, and then back off a bit. Don't try to get rid of all the noise -- it'll leave your dialogue sounding very empty, sterile, and artificial.

  • It is also important to keep the originals muted on an X track. Not doing so can get you in very big trouble on the stage. – Stavrosound Mar 24 '13 at 8:15

If you're the one doing the predubs/mix, check this out: http://www.stavrosound.com/blog/wordpress/2011/07/dialogue-clean-out-the-noise/

THis is effectively what a dub stage will do (with either C4, Dolby CAT 43, CEDAR, etc), only using something like RX or X-Noise on isolated individual basis

Routing is a must in my opinion, all DX edit tracks (EQ, trims, etc) -> DX sidechain (compressor, de-esser, noise suppressor) -> DX submaster (1:1 passthrough, maybe a brickwall limiter) -> DX Stem (print track)

(I'm glad the blog post has been helpful to people and share with them some insight into one noise suppression technique, I'm thankful for that, and I don't mean my answer to come across short or rudely succinct either, however I just feel this question has been asked here many times over, a search on SSD will reveal a lot of detailed answers)

Hope this helps and good luck!


Yes I have to mix as well! I have already read you article, was very useful, thank you, my question is more about the session setting and work flow. Assuming a tipical basic situation of indipendent movie with one boom directly in camera,  the pictire editor have finished his editing and sent to me one dialogue track OMF file. Now many scenes have on differents angles same ROOM TONE and I leave them on same track, other scenes have angles with very different RT and I split them om different tracks. In  all cases I first do the job of cleaning removing extra noises smoothing etc,  After that the 75% of scenes  have a loud broadband noise on them ( traffic, car noise hiss), so how do I route tracks and ho apply plug in, scene by scene on an aux  output, via audio suite region by region and render? Consider that unfortunaly times are short. 

Thanks for your patience. 

  • Well, it would be the signal flow I stated above. A series of recursive aux busses. in the track-level, use EQs on each track and take advantage of the "write to all enabled" function to apply custom EQ automation to each scene for corrective purposes. Then all of these tracks get summed through your sidechain aux, and this is where you run your compression and broadband noise suppression (post-EQ basically, and it's being performed on a summed signal). This summed signal is then outputted to your DX submaster. All of these plugns should be run as RTAS in this case. – Stavrosound Mar 24 '13 at 8:12

I am new at protools, and just exploring the snapshot function of automation, unfortunately does not works with pt 10 but I can in real time make and store any variation to any plug in parameter along the tracks it is applied, what is not clear to me is "Then all of these tracks get summed through your sidechain aux" , what do you mean for side chain, have used sidechain to drive for ex, a compressor on a kick drum with the bass guitar input, did you mean something like that or just submaster tracks to an aux stereo tracks and there applying all rtas plug in needed.

  • The "write to all enabled" now requires CPTK to add that functionality, in PT8 and below it only required DVTK. A sidechain is another (but probably more accurate) term for an aux bus. It's a 'chain' of plugins/processes happening in tandem, but not tied to, the signal flow going to the Mains (hence, 'side'). Think of a sidechain as a service road paralleling a freeway. You jump off, get gas use the bathroom etc, then you jump back on at the next on ramp. – Stavrosound Mar 25 '13 at 7:22
  • The signal flow being DX edit tracks (Audio tracks) bussed -> DX sidechain (1 Aux track, or 2 if you want to keep ADR on a separate signal path). Because all your DX edit trakcs have now been summed through your 1 Aux (the sidechain, or split between 2 if you keep ADR seperate), these Aux (sidechains) are bussed -> DX Submaster (Aux track). This serves as your master fader for what will be sent 1:1 to the DX Stem print track, and it allows you to place global metering plugins on DX if needed, as well as a brickwall limiter to ensure nothing exceeds QC spec when the stem gets printed. – Stavrosound Mar 25 '13 at 7:27
  • the DX STEM print track is an audio track where to record the final bounce? – alidav Mar 25 '13 at 7:40
  • yes, this sends 1:1 to the printmaster (another print) where the DX, FX, MX stems get summed. – Stavrosound Mar 25 '13 at 16:22
  • you usually only print the DX stem when you're done, or if you print along the way, you do so using Destructive Punch – Stavrosound Mar 25 '13 at 16:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.